CNS Core:Medium:Systems Challenges in Scaling Distributed Intelligent Applications

University Of Chicago, Chicago IL

Investigators

Abstract

The rapid proliferation of smart edge devices and breakthroughs in machine learning are poised to change how applications extract insights from sensor data, creating a new class of Distributed Intelligent Applications (DIAs) that include video monitoring, autonomous device control, and response for network cybersecurity. The project explores new analysis, distribution, and resource management approaches that enable DIAs to employ geographically distributed architectures, spreading analytics across sensors, edge clusters, and cloud datacenters. This approach is a major shift from today's approach that collocates data and analytics in the cloud. Distributed Intelligent Applications (DIAs) present fundamental challenges arising from distribution, bursts and high computational demands, as well as complex analytics. First, the project will explore systematic approaches that distribute DIA analysis across sensor, edge, and cloud resources to reduce cost and increase application quality. Second, the project blends analytics and communication, enabling new options to optimize capability and cost. Third, to meet realtime and bursty resource requirements, the project will explore new resource management approaches for shared, open resource environments that provide essential application guarantees. The proposed research will create capabilities that will increase the number and the capability of DIAs, allowing such applications greater accuracy, timeliness, and efficiency at lower cost. The project will involve a number of undergraduate and graduate students, exposing them to real edge computing testbeds, experimental studies, and exciting applications. These opportunities and impacts will include course enrichment and undergraduate research opportunities. The project team will bring novel insights and energy to several large outreach programs to broaden participation in computing. The project will release data and source code generated by the project. Such software artifacts will be open sourced via github (https://github.com/) or departmental storage service, for the project period. After the project's end, the team will provide best-effort support, helping others to use the project results as expertise and team effort is available. As opportunity arises, we will work with industry partners to make results and insights even more broadly available. This award reflects NSF's statutory mission and has been deemed worthy of support through evaluation using the Foundation's intellectual merit and broader impacts review criteria.